Christian, Perhaps you have been looking at Ole Husgaard, John Hodgkinson and Martin Boll's xti library.
Not to belittle their effort, but I found that libarary to not even be fork() safe far less thread-safe. There are many other problems with the library that make it unsuitable for use by more than a single process. (Maintaining per-stream state information in heap memory, holding copies of retrieved messages in heap memory, formulating messages and processing responses on the heap.) I also found some core XTI functionality missing (such as T_GODATA and T_GOEXDATA, t_sysconf(), atm extensions, etc.). Also, it installs as libnsl (which conflicts with Linux libnsl). It should likely install on Linux as libxnet (per XNS5 recommendations) even though it is part of libnsl on other OSs. I have written (but not tested) a libxnet library that will run on both LiS and Linux Fast-STREAMS (in fact because it uses standard facilities, it should run on other OSs). It was written as part of the OpenSS7 project. It is currently GPL, but I think you probably want LGPL. The library only holds read-only non-state getinfo information (data, address, option sizes) on the heap. It does not maintain TPI state information in the library by relies upon the underlying transport provider to maintain state. It can be made pthread safe (although it is only fork() safe at the moment). It does not keep copies of retreived messages on the heap, but uses zero wait polls and I_PEEK to avoid keeping messages on the heap. The OpenSS7 project, being an opensource project, is always in need (sometimes dire need) of funding. If you are interested in funding the completion of this library, contact me offline and I would be happy to discuss it. --brian +1 780 490 1141 On Wed, 08 Oct 2003, Christian Hildner wrote: > Hi, > > we need a reliable XTI networking interface for Linux. I have seen an > implementation, but that seems not to be nearly as robust as we know XTI > to be in other OSes like solaris, ... . > > If a member of this list is interested in the (further) developement of > XTI please contact me. We willing to give some payment for that work > although it would finally be GPL'd. > > If there is no interest from anybody: What are the key steps to > implement that interface if the following requirements must be met: > > - automatic cleanup of resources after termination of a process using XTI > - threadsafty > > All kinds of response are welcome. > > Christian > > > _______________________________________________ > Linux-streams mailing list > [EMAIL PROTECTED] > http://gsyc.escet.urjc.es/mailman/listinfo/linux-streams -- Brian F. G. Bidulock � The reasonable man adapts himself to the � [EMAIL PROTECTED] � world; the unreasonable one persists in � http://www.openss7.org/ � trying to adapt the world to himself. � � Therefore all progress depends on the � � unreasonable man. -- George Bernard Shaw � _______________________________________________ Linux-streams mailing list [EMAIL PROTECTED] http://gsyc.escet.urjc.es/mailman/listinfo/linux-streams
